Kapa's reranker benchmark: Voyage edges Jev on score, cost and latency

On company-knowledge retrieval, Voyage rerank-3 scored 0.535 at $1.91 per 1,000 questions and 0.18s median latency. A steerable general decision model, Jev 1.14, came within 0.003. Useful data for picking a reranker, but test on your own corpus.

• postVoyage rerank-3 models land on Vercel AI Gateway
• postVoyage embedding and reranking gain European data residency in Atlas
• postVoyage's rerankers start taking instructions at query time
• postWhy code retrieval benchmarks mislead, and how to rebuild them

• articleJev is the fastest-adopted model in AI Gateway historyAmelia Charles
• postCase Study: TypeSafe's Jev Model Beats Frontier LLMs on Classification TasksTypeSafe AI
• postArtificial Analysis launches Decision Index for classifier-style modelsArtificialAnlys

Checking sign-in…

Loading comments…